80 Vol. 1 FORESTS AND SCRUBLANDS OF NORTHERN FIORDLAND J. WARDLE, J. HAYWARD, and J. HERBERT, Forest and Range Experiment Station, New Zealand Forest Service, Rangiora (Received for publication 18 January 1971) ABSTRACT The composition and structure of the forests and scrublands of northern Fiordland were recorded at 1,053 sample points. The vegetation at each sample point was classified into one of 16 associations using a combination of Sorensen's 'k' index of similarity, and a multi-linkage cluster analysis. The associations were related to habitat and the distribution of each was determined. The influence of the introduced ungulates, red deer and wapiti, on the forests and scrublands was determined. Stand structure was analysed to provide infor­ mation on the susceptibility of the vegetation to damage from browsing and on the history of ungulate utilisation of the vegetation. Browse indices were calculated to provide information on current ungulate utilisation of the vegetation. INTRODUCTION A reconnaissance of northern Fiordland was carried out during the summer of 1969-70 by staff of the Forest and Range Experiment Station. The purpose was to describe the composition, structure, and habitat of the forest and scrub associations, to determine both present and past influence of ungulates on them, and to establish a number of permanent reference points to permit measurement of future changes in the vegetation. The area studied lies between the western shores of Lake Te Anau and the Tasman Sea. The southern boundary is the South Fiord of Lake Te Anau, the Esk Burn and Windward River catchments, and Charles Sound; the northern boundary is the Worsley and Transit River catchments (Fig. 1). The total area is ca. 700,000 acres of which almost two-thirds supports a forest or scrub cover. The area is a deeply dissected upland rising to over 6,000ft altitude in places. The highest peaks are the Llawrenny peaks (6,440ft) and Castle Mountain (6,872ft) in the north, and Mt Irene (6,085ft) and Mt Lyall (6,097ft) in the south. The main mountain ranges are the result of considerable late Tertiary and Pleistocene movement on more or less north-trending faults. The main alpine fault follows along the western limits of the survey area and forms a fault-line coast. Successive ice advances in the Pleistocene have extensively modified the landscape to give deeply eroded glacial valleys, fiords, lake basins, cirques (which may be lake filled), sharpened mountain peaks, and aggraded N.Z. JI For. Sci. 1 (1): 80-115 No. 1 Wardle, Hayward, & Herbert —• Forests of Northern Fiordland 81 £J FOREST & SCRUB PLOT LINES O OFF LINE PLOTS FIG. 1—Map of Survey area, northern Fiordland, showing line and plot positions. 82 New Zealand Journal of Forestry Science Vol. 1 main valleys with vast quantities of outwash alluvium (Fig. 2). The altitude of the cirque floors is commonly between 3,500ft and 4,000ft. FIG. 2—The main Worsley valley. Note the steep valley sides and wide terraces typical of glacial sculptured landscapes. The majority of the rocks are hard grey to pale green dioritic gneisses of the Bradshaw, Wet Jacket, Long Sound, and Milford Formations dating from the lower Permian and Cambrian. Sandstones, siltstones, and limestones occur near the shores of Lake Te Anau and are most common between the middle and south Fiords (Wood, 1960, 1962). The major forest soils are classified as Titiraurangi (N.Z. Soil Bureau, 1968). These are thin stony loams and sandy loams developed on schist, granite, diorite, and Tertiary rocks and are of very low fertility. The Titiraurangi soils may give way to soils of the McKerrow-Resolution series under subalpine scrub. Both soils range from a fibrous peat to a peaty loam overlying a sand or sandy loam over rock. The area lies within the zone of prevailing westerly winds. Moisture laden winds from the Tasman Sea cause heavy precipitation and frequent mist, especially in the west, when they are forced to rise over the mountain ranges. Precipitation is heavy throughout the year but tends to be greatest in the summer months. In the winter much of the precipitation above 3,000ft falls as snow. There are no climate stations or even isolated rain gauges in the survey area. The nearest climate records are from Milford Sound to the north, Te Anau to the east, and Doubtful Sound and Wilmott Pass to the south. Milford Sound and Doubtful Sound have long-term average annual rainfalls of No. 1 Wardle, Hayward, & Herbert — Forests of Northern Fiordland 83 253in. and 226in. respectively. Wilmott Pass has 28lin. annually but at Te Anau the rainfall is only around 50in. (N.Z. Met. Service, 1969). It can be assumed that the rainfall of the survey area is more than 200in. west of the Main Divide but drops towards the south-east, until, in the extreme east of the Murchison Range it is probably less than 80in. The botany and ecology of the forests and scrublands of the area have never been satisfactorily described. The most thorough account to date is published in the report of the New Zealand-American Fiordland Expedition (Poole, et al. 1951). This account describes the important forest and scrub types between George Sound, Caswell Sound, and the Middle Fiord of Lake Te Anau and highlights some of the more interesting features of the vegetation-habitat relationships of the area such as the unusual plant succession on the steep glaciated valley sides. Forest succession on landslides above Lake Thomson, at the head of the north-west arm of the Middle Fiord of Lake Te Anau has been further described by Mark et al. 1964. Scott et al. 1964 have described altitudinal variation in forest composition near Lake Hankinson, but the vegetation outside this George Sound-Caswell Sound-Middle Fiord area has never been fully studied, and only broad accounts of it exist (Holloway 1954, Cockayne 1928). The following account of the animal history of the survey area is from K. Tustin (pers. comm.). The first herbivorous mammal to be introduced into the area was the opossum {Trichosurus vulpecula) in the early 1890s. Subsequent liberations up to 1930 have resulted in a local opossum population at Milford Sound, and light but more widespread populations around the western shores of Lake Te Anau. Wapiti (Cervus canadensis) were liberated in George Sound in 1905, and red deer {Cervus elaphus) infiltrated into the area, mostly from the south, and became widespread by the 1950s. Most of the survey area could be considered to have experienced peak populations during the 1940s to early 1960s, with the central survey areas first affected. The north­ western area is an exception and is still in the stages of colonisation. Scattered reports of chamois {Rupicapra rupicapra) throughout the area have been reported and some hare (Lepis europaeus) signs seen but these animals are, as yet, of little significance in the animal-plant interrelationships. TECHNIQUE Field Measurements Sociological descriptions of 1,053 stands throughout the area, form the basis of the present report. A further 25 permanent plots will form datum points to aid interpretation of future changes which may occur in the vegetation. The layout and measurement of these permanent plots will be described elsewhere. Nine hundred and thirteen of the sociological descriptions or temporary plots were located at 147.6ft (135m) intervals along 84 alitudinal transects. The starting points of these transects were chosen in a restricted random fashion along the major stream beds. In each case the direction of the transea followed the compass bearing representing the shortest distance plus 5° from the random starting point to the top of the scrub belt (Fig. 1). The remaining 140 temporary plots were located in off-line positions. These were subjectively 84 New Zealand Journal of Forestry Science Vol. 1 placed in representative stands of minor forest and scrub associations not sampled, or sampled only rarely by the on-line plots in an area. Five tiers of vegetation were delineated by these heights: stand top height, 40ft, 15ft, 6ft, 1ft, and ground level. All species of vascular plants in each tier were listed; epiphytes and parasites were recorded separately. An estimate was made of the physiognomically dominant species and of the density of vegetation in each tier. The aspect, altitude, slope, physiography, parent rock, soil drainage, and mean soil depth (measured with metal probes) at the site of each temporary plot were recorded. The percentages of soil surface covered by moss, litter, living vascular plant material, bare soil and rock were estimated. Rock material visible on the soil surface was recorded as being loose rock or parent rock and an estimate made of whether loose rock particles were predominantly greater than or less than 1ft in diameter. Species of vascular plants which showed obvious ungulate browse were noted and the degree of browse recorded as light, moderate, or heavy. Analysis (a) Associations The 1,053 temporary plots were divided into 16 associations using the numerical procedure described below. Two hundred plots were randomly selected from the total and an index of similarity, "Sorensen's k" was calculated between each pair of plots. This involved assigning each species in each plot an "importance rating" based on physiognomic dominance. Physiognomic dominant species were rated 100 points, other species 50 points, and comparisons were made on the basis on importance ratings shared, over total importance ratings. The highest indices of similarity were progressively clustered using a multilinkage technique (Wardle, 1970) until 16 groups of more than three plots each remained. The mean importance rating for each species in each of these groups was calculated and all 1,053 plots were then tested against the 16 groups, again using "Sorensen's k"*.